Cambridge University Press 0521609186 - Double Affine Hecke Algebras Ivan Cherednik Excerpt More information Chapter 0 Introduction 0.0 Universality of Hecke algebras 0.0.1 Real and imaginary Before a systematic exposition, I will try to outline the connections of the representation theory of Lie groups, Lie algebras, and Kac–Moody algebras with the Hecke algebras and the Macdonald theory. The development of mathematics may be illustrated by Figure 0.1. Imaginary axis (conceptual mathematics) Real axis (special functions, numbers) Figure 0.1: Real and Imaginary Mathematics is fast in the imaginary (conceptual) direction but, generally, slow in the real direction (fundamental objects). Mainly I mean modern mathematics, but it may be more universal. For instance, the ancient Greeks created a highly conceptual axiomatic geometry with modest “real output.” I do not think that the ratio / is much higher now. 1 © Cambridge University Press www.cambridge.org Cambridge University Press 0521609186 - Double Affine Hecke Algebras Ivan Cherednik Excerpt More information 2 CHAPTER 0. INTRODUCTION Let us try to project representation theory on the real axis. In Figure 0.2 we focus on Lie groups, Lie algebras, and Kac–Moody algebras, omitting the arithmetic direction (ad`elesand automorphic forms). The theory of special functions, arithmetic, and related combinatorics are the classical objectives of representation theory. Im Representation theory of Lie groups, Lie algebras, and Kac-Moody algebras Re 1 Spherical functions 2[Characters of KM 3 Vλ ⊗ Vµ : Vν]4[Mλ : Lµ] algebras (irreps of dimC < ∞) (induced: irreps) Figure 0.2: Representation Theory Without going into detail and giving exact references, the following are brief explanations. (1) I mean the zonal spherical functions on K\G/K for maximal compact K in a semisimple Lie group G. The modern theory was started by Berezin, Gelfand, and others in the early 1950s and then developed significantly by Harish-Chandra [HC]. Lie groups greatly helped to make the classical theory multidimensional, although they did not prove to be very useful for the hypergeometric function. (2) The characters of Kac–Moody (KM) algebras are not far from the prod- ucts of classical one-dimensional θ–functions and can be introduced without representation theory (Looijenga, Kac, Saito). See [Lo]. How- ever, it is a new and important class of special functions with various applications. Representation theory explains some of their properties, but not all. (3) This arrow gives many combinatorial formulas. Decomposing tensor products of finite dimensional representations of compact Lie groups and related problems were the focus of representation theory in the 1970s and early 1980s. They are still important, but representation theory moved toward infinite dimensional objects. (4) Calculating the multiplicities of irreducible representations of Lie alge- bras in the BGG–Verma modules or other induced representations be- longs to conceptual mathematics. The Verma modules were designed © Cambridge University Press www.cambridge.org Cambridge University Press 0521609186 - Double Affine Hecke Algebras Ivan Cherednik Excerpt More information 0.0. UNIVERSALITY OF HECKE ALGEBRAS 3 as a technical tool for the Weyl character formula (which is “real”). It took time to understand that these multiplicities are “real” too, with strong analytic and modular aspects. 0.0.2 New vintage Figure 0.3 is an update of Figure 0.2. We add the results which were obtained in the 1980s and 1990s, inspired mainly by a breakthrough in mathematical physics, although mathematicians had their own strong reasons to study ge- neralized hypergeometric functions and modular representations. Im Representation theory Re 1 Spherical fns23 KM characters[Vλ ⊗ Vµ : Vν ] 4 [Mλ : Lµ] e Generalized e e 1 2 Conformal 3 Verlinde e4 Modular hypergeom. blocks algebras functions reps Figure 0.3: New Vintage (1˜) These functions will be the main subject of the first chapter. We will study them in the differential and difference cases. The interpretation and generalization of the hypergeometric functions via representation theory was an important problem of the so-called Gelfand program and remained unsolved for almost three decades. (2˜) Actually, the conformal blocks belong to the (conceptual) imaginary axis as well as their kin, the τ–function. However, they extend the hyperge- ometric functions, theta functions, and Selberg’s integrals. They attach the hypergeometric function to representation theory, but affine Hecke algebras serve this purpose better. (3˜) The Verlinde algebras were born from conformal field theory. They are formed by integrable representations of Kac–Moody algebras of a given level with “fusion” instead of tensoring. These algebras can be also © Cambridge University Press www.cambridge.org Cambridge University Press 0521609186 - Double Affine Hecke Algebras Ivan Cherednik Excerpt More information 4 CHAPTER 0. INTRODUCTION defined using quantum groups at roots of unity and interpreted via operator algebras. (4˜) Whatever you may think about the “reality” of [Mλ : Lµ], these multi- plicities are connected with the representations of Lie groups and Weyl groups over finite fields (modular representations). Nothing can be more real than finite fields! 0.0.3 Hecke algebras The Hecke operators and later the Hecke algebras were introduced in the the- ory of modular forms, actually in the theory of GL2 over the p–adic numbers. In spite of their p–adic origin, they appeared to be directly connected with the K–theory of the complex flag varieties [KL1] and, more recently, with the Harish-Chandra theory. It suggests that finite and p–adic fields are of greater fundamental importance for mathematics and physics than we think. Concerning the great potential of p–adics, let me mention the following three well-known confirmations: (i) The Kubota–Leopold p–adic zeta function, which is a p–adic analytic continuation of the values of the classical Riemann zeta function at negative integers. (ii) My theorem about “switching local invariants” based on the p–adic uni- formization (Tate–Mumford) of the modular curves which come from the quaternion algebras. (iii) The theory of p–adic strings due to Witten, which is based on the sim- ilarity of the Frobenius automorphism in arithmetic to the Dirac operator. Observation. The real projection of representation theory goes through Hecke-type algebras. The arrows in Figure 0.4 are as follows. (a) This arrow is the most recognized now. Quite a few aspects of the Harish- Chandra theory in the zonal case were covered by representation theory of the degenerate (graded) affine Hecke algebras, introduced in [Lus1] ([Dr1] for GLn). The radial parts of the invariant differential operators on symmetric spaces, the hypergeometric functions and their general- izations arise directly from these algebras [C10]. The difference theory appeared even more promising. It was demon- strated in [C19] that the q–Fourier transform is self-dual like the classi- cal Fourier and Hankel transforms, but not the Harish-Chandra trans- form. There are connections with the quantum groups and quantum © Cambridge University Press www.cambridge.org Cambridge University Press 0521609186 - Double Affine Hecke Algebras Ivan Cherednik Excerpt More information 0.0. UNIVERSALITY OF HECKE ALGEBRAS 5 Representation theory Im Representation theory of Hecke algebras Kazhdan-Lusztig Macdonald theory, double Hecke algebras polynomials ˜b ! ?! a˜ b ? ca d˜ d ? ˜ c Re 1 Spherical fns 2 KM-characters 3 [Vλ ⊗ Vµ : Vν ] 4 [Mλ : Lµ] 1˜ 2˜ ConformalHypergeom. 3˜ Verlinde 4˜ Modular fns blocks algebras reps Figure 0.4: Hecke Algebras symmetric spaces (Noumi, Olshansky, and others; see [No1]). However, the double Hecke algebra technique is simpler and more powerful. (b) The conformal blocks are solutions of the KZ–Bernard equation (KZB). The double Hecke algebras lead to certain elliptic generalizations of the Macdonald polynomials [C17, C18, C23] (other approaches are in [EK1, C17, FV3], and the recent [Ra]). These algebras govern the monodromy of the KZB equation and “elliptic” Dunkl operators (Kirillov Jr., Felder– Tarasov–Varchenko, and the author). The monodromy map is the inverse of arrow (˜b). The simplest examples are directly related to the Macdonald polynomials and those at roots of unity. (c) Hecke algebras and their affine generalizations give a new approach to the classical combinatorics, including the characters of the compact Lie groups. The natural setting here is the theory of the Macdonald polynomials, although the analytic theory seems more challenging. Concerning (˜c), the Macdonald polynomials at the roots of unity give a simple approach to the Verlinde algebras [Ki1, C19, C20]. The use of the nonsymmetric Macdonald polynomials here is an important de- © Cambridge University Press www.cambridge.org Cambridge University Press 0521609186 - Double Affine Hecke Algebras Ivan Cherednik Excerpt More information 6 CHAPTER 0. INTRODUCTION velopment. Generally, these polynomials are beyond the Lie and Kac– Moody theory, although they are connected with the Heisenberg–Weyl and p–adic Hecke algebras. (d) This arrow is the Kazhdan–Lusztig conjecture proved by Brylinski–Kashi- wara and Beilinson–Bernstein and then generalized to the Kac–Moody case by Kashiwara–Tanisaki. By (d˜), I mean the modular Lusztig conjecture (partially) proved by Anderson, Jantzen, and Soergel. There is recent significant progress due to Bezrukavnikov. The arrow from the Macdonald theory to modular representations